Re: blocked libtool
+ Thorsten Alteholz (Wed, 06 May 2009 13:22:53 +0200):
> Hi Adeodato,
Hello, Thorsten (hope it's okay I'm quoting you in public).
> could you please tell me the reason for blocking libtool's transition
> from unstable to testing? I have a few packages that depend on libtool
> but don't want to disturb debian-release if there is still a reason for
> blocking it.
Thorsten asks about this output in the migration pages:
| Trying to update libtool from 1.5.26-4 to 2.2.6a-4 (candidate is 28 days old)
| Not touching package, as requested by adeodato (contact debian-release if update is needed)
I thought maybe more people are wondering about this, and I added an
explanatory paragraph on my hint file [1], which I reproduce now:
# == Performance blocks ==
# Block here some transitions so that britney runs faster. Even if any
# of these packages is a candidate for migration itself (older than 10
# days, etc.), it doesn't mean it will be able to migrate, since all of
# its reverse dependencies have to be ready as well. But britney chokes
# rather badly on some of these packages, taking a lot of time to
# process them and their reverse dependencies; because of this, and to
# keep ftp-master free of spurious CPU churn (it's a host that sees
# quite a lot of interactive use), we prevent britney from trying to
# migrate some packages for as long as it wouldn't succeed anyway. (This
# is determined by a human, but in general a package being blocked out
# of this paragraph implies there's nothing else preventing migration
# than waiting for the reverse dependencies to be ready.)
block libtool
block gnome-sharp2
block totem-pl-parser
FWIW, this is only done for transitions that are big enough as to cause
a noticeable slow down (libtool eg. involves around 600 Bin-NMUs), or
transitions that britney really chokes on when calculating uninstallabilities
(eg. totem-pl-parser, no idea why).
Cheers,
[1]: http://release.debian.org/britney/hints/adeodato
--
- Are you sure we're good?
- Always.
-- Rory and Lorelai
Reply to: